Alan Beavitt, Scorraig, Dundonnell, Ross-shire, Scotland
Introduction
If a precise copy is made of a famous old violin, the copy will not sound the same as the original. The best modern makers compensate for an apparent difference in the stiffness of new wood by making their instruments slightly thicker. (Saunders1) The difference in the quality of the wood is usually attributed to some secret process known only to the old masters, and numerous proposals or guesses have been made over the years. Sacchoni2 for example, sensing that the wood of old violins was somehow stiffer, advocated the use of silicate solutions to "ossify" the wood. Other wood treatments that have been suggested include baking, boiling, soaking in sea water and infection with moulds. More recently, Yano and Minato3 have used a vapour treatment to improve the accoustic properties.
It is generally recognised that the sound of a new violin changes with time. The effect is known as "playing-in", but does the instrument actually have to be played? It is the thesis of this article that the change in sound of a new violin, and the apparent difference between old and new violins is mainly due to the effects of variations in atmospheric humidity. A conditioning process is described, which simulates these variations and allows the maker to produce violins of lasting excellence.
As a player as well as a maker, I had noted the change in sound of new violins (throughout this article the word 'violin' is taken to refer to other members of the violin family also) hanging on the workshop wall - and only played occasionally - so that the change could not be due to the effect of vibrating strings.
Wood is a rheological substance - it not only responds to a stress with a reversible instantaneous movement, but there is also an additional gradual movement known as creep. It occurred to me that the change in sound of the violin was consistent with an increase in stiffness which might be caused by creep.
To make a measurement of stiffness, a sensitive dial indicator (Baty lever type, 0-0.2 mm) was used to note the deflection of the bridge due to an additional load or due to one or more of the strings being released. The elastic properties of wood vary with temperature and moisture content; in a violin the wood loses and gains moisture according to the humidity of the air circulating via the "f" holes. The stiffness measurements were therefore made at constant humidity and temperature.
It was found that there is indeed an increase of about 10% in the stiffness of a new violin, and with the stiffness increasing with a time-constant of about one week. The change was matched by a decrease in bridge height, suggesting that creep was responsible.
That creep was over in a matter of weeks surprised me at first, because to my ear the violin sound was still changing after six months. Also, I wondered how an old violin could creep into the distorted shape shown in figure 1.
The answer I believe lies in humidity cycling. The stiffness measurements were made at constant humidity whereas atmospheric humidity is constantly changing. Hearmon and Paton4 found that enhanced creep was found in small samples of stressed wood subjected to humidity cycling. It is the unusual combination of high stress and thin section (allowing for a relatively rapid adsorption of moisture) that makes the effect significant in a violin. These authors also reported that the enhanced creep effect was linear - in other words, a few large-amplitude humidity cycles have the same effect as many small ones. Thus we can simulate the effect of a long period of atmospheric humidity variation in a relatively short interval.
Measurement of the Effects of Humidity Cycling
Two methods can be used to take a violin through a humidity cycle. In the first, a moisture proof box contains a tray of a saturated salt solution and a miniature fan blowing into one of the "f" holes. (Varnish acts as a moisture barrier, so the only effective way in is through the "f" holes.) Saturated solutions (the solution contains undissolved salt) of various salts have the property of maintaining a constant humidity in an enclosed space (the values are listed in reference 5). Potassium nitrate solution (giving a relative humidity of 93%) was used for the wetting part of the cycle and potassium carbonate (with a published value of 44% but a hygrometer reading of 63%) for the drying. To effect a humidity cycle the salt is changed over after a suitable period. Violins were not cycled below normal humidity because of the risk of cracking.
The second method (fig. 2) uses a small air pump, as used for aerating an aquarium but with a modified inlet port. The pump circulates air through a container (a one gallon demi-john is ideal) of saturated salt solution, through the violin and back through the pump. Masking tape (test on an inconspicuous patch of varnish first!) can be used to seal off the "f" holes around the inlet and outlet tubes.
The second method is preferable for the wetting cycle as it avoids the constant need to adjust the pegs (they get too tight in the wetting cycle and too loose in the drying)
The uptake and loss of moisture in a violin can be monitored by weighing. The equipment needed to weigh a 400 gram object with sufficient accuracy is expensive. However, a simple beam balance with string suspension (fig. 3) can be used to monitor the weight change, using coins as weights (1p weighs 3.56g). Although easy to make, it can nevertheless measure a weight change of 0.1g.
Fig. 3. A simple beam balance for measuring the weight change of a violin.
It can be seen from fig. 4 that it takes 5 days for the violin to come into near equilibrium with a moisture change. (The curve includes a contribution by the corner and end blocks, which take longer, but are of no concern here).
The uptake of moisture can also be observed by surface appearance. High moisture content causes the wood grain to swell, and the effect gives the varnish surface a texture. During the first wetting cycle, the effect can be observed in one day at the thinner parts of the back, but it takes four days to show in the thicker part at the centre of the back.
Creep can be monitored with a modified micrometer (fig.5) which measures the bridge height relative to the ends of the body.
Fig. 5. Modified micrometer for monitoring creep.
The effect of humidity cycling on creep is shown in fig. 6. These measurements were made on a new violin that had been strung up for three weeks and was therefore past the initial settling-in period. The bridge rises with an increase in humidity but the rise is masked by creep. Creep generally causes the bridge height to decrease, but as illustrated by fig. 6, there may be episodes when it rises also.
To the ear, the effect of humidity cycling is quite striking, especially on the higher strings, which become more sonorous and resonant. There does not appear to be further improvement after six cycles, so the process, which I call the conditioning process, is deemed to be complete (the entire process taking sixty days).
To get an objective view of the effects of humidity cycling I constructed a bowing maching with a reciprocating motion driven by an electric motor. By this means, a reproducible value of bow speed, pressure and distance from the bridge could be obtained. Sound from the four open strings was analysed harmonically and the preliminary results are shown in fig. 7. These results are from a viola, before and after only two humidity cycles. Measurements are continuing and will be reported at a later date.
To summarize the results of harmonic analysis so far, the ratio between the fundamental and the sum of all the harmonics is unchanged, but there is a reduction in the harmonic content above 4 kHz, and a reduction in the anharmonic content especially on the A string. The anharmonics are the peaks in the readout that are not at regular intervals, and detract from sonority. The ear appears to be very sensitive to the anharmonic content of violin tone. Previous research (eg. Meinel6, Saunders1) has concentrated on the harmonic content and this may explain why players sometimes claim that there is a difference between violin A and violin B while the researchers cannot find one.
To get the opinion of professional musicians, I asked some of the staff of the Royal Northern College of Music to play a violin that had been through the conditioning process but had never been played, except for setting up.
"I felt there was a desirable evenness across all four strings... often with a new instrument and particularly one that has not been played you encounter a weaker range or tonal area. The G string had plenty of depth, there was a special quality high on the middle strings and the E string had both power and quality. What impressed most was the clarity of the instrument and how well it spoke even at high speeds. Most of all I wanted to carry on playing it - always a good sign - and try it in a bigger hall." - Christopher Rowland.
"There is something very rounded and mature in the sound that belies its new and unplayed nature. I was also fortunate to hear the test made by Christopher Rowland and was again impressed by the power, clarity and warmth of the sound across its whole range" - Roger Bigley
A 'before and after' opinion on one of my violins made in 1991, repaired and revarnished in 1995 and conditioned six months later:
"Whatever you did to my violin, it is hugely more resonant. It has a creamy tone particularly on the G string, the whole violin has eveness, immediacy and speaks clearly, especially on the E string" Robert Riley
One possible explanation for the effect on sound of the conditioning process lies in the re-distribution of stress. In a violin the highly stressed parts are around the sound post and at each end of the body, and these parts will creep preferentially, re-distributing stress. From the principle of superposition this would not be expected to change the sound if the violin were made out of an isotropic substance, such as glass, but might do so in wood which is highly anisotropic (the elastic constants are different along the three axes)
Another explanation concerns changes in the wood itself. If the experimenter monitors the weight change of a violin during humidity cycling, as in fig. 4, it will be noted that the final weight is not necessarily the same as the starting weight during the first cycle. It is one of the odd properties of wood that the equilibrium moisture content depends on whether the wood is gaining or losing moisture (as shown in fig. 8). It can be seen that the curves for adsorption and desorption move closer together after the second and subsequent cycles, so humidity cycling must cause a chemical or physical change to the wood. From the measurements of Noak and Becker7 it is known that moisture content has a large effect on the damping loss (the rate at which a vibrating body loses energy) of wood, so a link between humidity cycling and damping (and therefore resonance) would not be surprising.
I guess that conditioning is not a magic process for turning a frog violin into a prince. If a violin is built like a trade fiddle it is likely to sound like a nineteenth century trade fiddle after conditioning.
The essential point is that the effects of humidity cycling will happen anyway. But the process allows the maker to achieve optimum thicknessing for the long-term and to make final adjustments to bridge, sound-post and bass-bar before handing over to the player.
It is appropriate to think partly in terms of evolution, when considering the tonal properties of old violins. (It must be remembered that they were originally baroque). The unknown restorers who have brought the best of the antique violins to their present state of perfection have had the advantage of building on firm foundations. The modern makers - until now - have had the problem of building on shifting sands, and of trying to predict the likely sound of their product in the future.
There is nothing un-natural about the conditioning process. The same effect could be obtained by travelling repeatedly between Arizona and Zaire, staying a few weeks at each location, without air-conditioning, and with the violin case open. The workshop method saves on time, trouble and aviation fuel.
The conditioning process has an incidental visual advantage. In fig. 5 it can be seen that the front develops an attractive reed-like texture, an effect which is accentuated by consolidation of the varnish (which may take several years. and have an effect on sound - Minato 8). As shown in fig. 9, the back develops the texture and wavy surface of the underlying figure often seen in old violins. These effects are probably due to the relief of internal stress.
Fig. 9. Showing the effect of the conditioning process on the figure of the back of a violin.
The violin maker should note that fingerboards and roughed-out neck blocks should always be exposed to humidity cycles (keeping them in the kitchen is helpful!) before final shaping. Seasoning is no guarantee of stability - I once made a neck from a sixty year old neck block which went on to develop a twist. The neck should be set initially too high to allow for the subsequent drop due to creep.
It is a popular concept of "playing-in" that vibrations somehow affect the structure of wood, but I believe the real cause of the effect is humidity cycling. It is noted that the (humid) breath of violin and viola players is in direct line with the "f" holes, giving a humidity cycle every time they play.
Finally, violinists claim that famous instruments kept in museum cases lose their quality. We have seen that wood is not the stable and predictable substance that it appears to be. We have the intriguing possibility that a violin needs some humidity cycling to maintain its quality.
Long term changes in the sound of violins are mainly due to variations in atmospheric humidity. A conditioning process has been devised which simulates these atmospheric variations and has the following effects:-
• Enhances appearance.
• Has a stabilising effect on dimensions critical to playing.
• Allows the maker to achieve optimum thicknessing for the long-term, and to make final adjustments to bridge, sound-post and bass-bar before handing over to the player.
• Improves sonority and resonance.
1 Saunders F., J. Acoust. Soc. Am., 17, 169-186
2 Sacchoni S., Secrets of Stradivari, Libreria del Convegno, Cremona 1979
3 Yano H. & Minato K., J. Acoust. Soc. Am., 92, 1222-1227
4 Hearmon R. & Paton J., For. Prod. J., 14 (8), 357-359
5 Kaye G. & Laby T., Tables of Physical and Chemical Constants, Longman 1986.
6 Meinel G., Sov. Phys. Acoust., 6, 149-161
7 Noak D & Becker H., Wood Science & Technology, 2, 213-230
8 Minato K. et al., Holzforschung, 49, 222-226
Acknowledgements
The author would like to thank Prof. Rodney Latham, University of Aston for much encouragement, and Dr. Murray Campbell, University of Edinburgh for the harmonic analysis.
Biography
Alan Beavitt (b. 1939) made his first violin while still at school, but was advised to follow the safer career of science. He worked as a research physicist in Australia and UK, moving to the Highlands of Scotland to become a postman and part time violin maker in 1974. Has been a full-time professional maker since 1982, specialising in Baroque and Modern violins and violas. He won the Facta Britannia violin-making competition in 1986.
